Comparing the performance of Matpower with our Polar ACOPF model (both using IPOPT with linear solver MA27) with standard options, we find that our model, although slightly slower, has very comparable time performance statistics to Matpower. The table below displays the real time (in seconds) it takes to converge to objective values within numerical tolerances of each other.
Note that our AC models are able to solve the 3375 bus Polish case, a case for which Matpower (on its basic settings) was unable to find a solution. In addition, the ACOPF models returned a better objective value in the 2383 bus Polish case. We believe this is due to the handling of symmetry in line definition, as case2383wp has both lines [344-346] and [346-344] in the dataset (as opposed to defining line [344-346] twice). The documentation on data utilities provides further information on how this is handled in data processing, such to avoid conflict when using the model archive.
| Test Case | Objective | Matpower Time | GAMS time |
| IEEE 6 Bus | 3.143974529e+03 | 0.1326 | 0.379 |
| IEEE 9 Bus | 1.254055725e+06 | 0.1429 | 0.222 |
| IEEE 14 Bus | 8.081526256e+03 | 0.1606 | 0.217 |
| IEEE 24 Bus | 6.335220252e+04 | 0.1989 | 0.255 |
| IEEE 30 Bus | 5.768923368e+02 | 0.1831 | 0.270 |
| IEEE 39 Bus | 4.186417779e+04 | 0.2440 | 0.281 |
| IEEE 57 Bus | 4.173778629e+04 | 0.2087 | 0.324 |
| IEEE 118 Bus | 1.296606850e+05 | 0.3419 | 0.397 |
| IEEE 300 Bus | 7.197250765e+05 | 0.6823 | 0.726 |
| Polish 2383wp* | 1.8630702523e+06 | 4.8205 | 6.006 |
| Polish 2736sp | 1.307883126e+06 | 5.2036 | 5.708 |
| Polish 2737sop | 7.776293002e+05 | 4.6181 | 5.740 |
| Polish 2746wop | 1.208279810e+06 | 4.6862 | 6.352 |
| Polish 2746wp | 1.631775103e+06 | 6.0592 | 6.640 |
| Polish 3012wp | 2.591706565e+06 | 7.2076 | 7.435 |
| Polish 3120sp | 2.142703764e+06 | 7.1659 | 7.809 |
| Polish 3375wp** | 7.412030674e+06* | 180.24* | 9.389 |
* Matpower converges to a different solution, 1.868511825e+06
** Matpower failed to solve this case within the time limit